#ifdef HAVE_DIX_CONFIG_H
#include <dix-config.h>
#endif
#include <stdlib.h>
#include <X11/X.h>
#include <X11/Xmd.h>
#include <X11/Xproto.h>
#include "gcstruct.h"
#include "windowstr.h"
#include "scrnintstr.h"
#include "pixmapstr.h"
#include "regionstr.h"
#include "cfb.h"
#include "cfbmskbits.h"
#include "cfb8bit.h"
#include "fastblt.h"
#include "mergerop.h"
#ifdef notdef
#if defined (FAST_UNALIGNED_READS) && PSZ == 8
#define DO_UNALIGNED_BITBLT
#endif
#endif
#if defined(FAST_MEMCPY) && (MROP == Mcopy) && PSZ == 8
#define DO_MEMCPY
#endif
#if PSZ == 24 && (MROP) == 0
#endif
#if PSZ == 24
#define BYPP 3
#if PGSZ == 32
#define P3W 4
#define PAM 3
#define P2WSH 2
#else
#define P3W 8
#define PAM 7
#define P2WSH 3
#endif
#endif
void
MROP_NAME(cfbDoBitblt)(
DrawablePtr pSrc,
DrawablePtr pDst,
int alu,
RegionPtr prgnDst,
DDXPointPtr pptSrc,
unsigned long planemask)
{
CfbBits *psrcBase, *pdstBase;
int widthSrc, widthDst;
BoxPtr pbox;
int nbox;
BoxPtr pboxTmp, pboxNext, pboxBase, pboxNew1, pboxNew2;
DDXPointPtr pptTmp, pptNew1, pptNew2;
int w, h;
int xdir;
int ydir;
CfbBits *psrcLine, *pdstLine;
register CfbBits *psrc;
register CfbBits *pdst;
MROP_DECLARE_REG()
CfbBits startmask, endmask;
int nlMiddle;
int xoffSrc, xoffDst;
register int nl;
int careful;
#if (PSZ != 24) || (MROP != 0)
register int leftShift, rightShift;
register CfbBits bits;
register CfbBits bits1;
#endif
#if PSZ == 24
#ifdef DO_MEMCPY
int w2;
#endif
#if MROP == 0
int widthSrcBytes = cfbGetByteWidth(pSrc);
int widthDstBytes = cfbGetByteWidth(pDst);
#endif
#endif
MROP_INITIALIZE(alu,planemask)
cfbGetLongWidthAndPointer (pSrc, widthSrc, psrcBase)
cfbGetLongWidthAndPointer (pDst, widthDst, pdstBase)
careful = ((pSrc == pDst) ||
((pSrc->type == DRAWABLE_WINDOW) &&
(pDst->type == DRAWABLE_WINDOW)));
pbox = REGION_RECTS(prgnDst);
nbox = REGION_NUM_RECTS(prgnDst);
pboxNew1 = NULL;
pptNew1 = NULL;
pboxNew2 = NULL;
pptNew2 = NULL;
if (careful && (pptSrc->y < pbox->y1))
{
ydir = -1;
widthSrc = -widthSrc;
widthDst = -widthDst;
if (nbox > 1)
{
pboxNew1 = (BoxPtr)ALLOCATE_LOCAL(sizeof(BoxRec) * nbox);
if(!pboxNew1)
return;
pptNew1 = (DDXPointPtr)ALLOCATE_LOCAL(sizeof(DDXPointRec) * nbox);
if(!pptNew1)
{
DEALLOCATE_LOCAL(pboxNew1);
return;
}
pboxBase = pboxNext = pbox+nbox-1;
while (pboxBase >= pbox)
{
while ((pboxNext >= pbox) &&
(pboxBase->y1 == pboxNext->y1))
pboxNext--;
pboxTmp = pboxNext+1;
pptTmp = pptSrc + (pboxTmp - pbox);
while (pboxTmp <= pboxBase)
{
*pboxNew1++ = *pboxTmp++;
*pptNew1++ = *pptTmp++;
}
pboxBase = pboxNext;
}
pboxNew1 -= nbox;
pbox = pboxNew1;
pptNew1 -= nbox;
pptSrc = pptNew1;
}
}
else
{
ydir = 1;
}
if (careful && (pptSrc->x < pbox->x1))
{
xdir = -1;
if (nbox > 1)
{
pboxNew2 = (BoxPtr)ALLOCATE_LOCAL(sizeof(BoxRec) * nbox);
pptNew2 = (DDXPointPtr)ALLOCATE_LOCAL(sizeof(DDXPointRec) * nbox);
if(!pboxNew2 || !pptNew2)
{
if (pptNew2) DEALLOCATE_LOCAL(pptNew2);
if (pboxNew2) DEALLOCATE_LOCAL(pboxNew2);
if (pboxNew1)
{
DEALLOCATE_LOCAL(pptNew1);
DEALLOCATE_LOCAL(pboxNew1);
}
return;
}
pboxBase = pboxNext = pbox;
while (pboxBase < pbox+nbox)
{
while ((pboxNext < pbox+nbox) &&
(pboxNext->y1 == pboxBase->y1))
pboxNext++;
pboxTmp = pboxNext;
pptTmp = pptSrc + (pboxTmp - pbox);
while (pboxTmp != pboxBase)
{
*pboxNew2++ = *--pboxTmp;
*pptNew2++ = *--pptTmp;
}
pboxBase = pboxNext;
}
pboxNew2 -= nbox;
pbox = pboxNew2;
pptNew2 -= nbox;
pptSrc = pptNew2;
}
}
else
{
xdir = 1;
}
while(nbox--)
{
w = pbox->x2 - pbox->x1;
h = pbox->y2 - pbox->y1;
#if PSZ == 24
#ifdef DO_MEMCPY
w2 = w * BYPP;
#endif
#endif
if (ydir == -1)
{
psrcLine = psrcBase + ((pptSrc->y+h-1) * -widthSrc);
pdstLine = pdstBase + ((pbox->y2-1) * -widthDst);
}
else
{
psrcLine = psrcBase + (pptSrc->y * widthSrc);
pdstLine = pdstBase + (pbox->y1 * widthDst);
}
#if PSZ == 24
if (w == 1 && ((pbox->x1 & PAM) == 0 || (pbox->x1 & PAM) == PAM))
#else
if ((pbox->x1 & PIM) + w <= PPW)
#endif
{
maskpartialbits (pbox->x1, w, endmask);
startmask = 0;
nlMiddle = 0;
}
else
{
maskbits(pbox->x1, w, startmask, endmask, nlMiddle);
}
#if PSZ == 24
#if 0
nlMiddle = w - (pbox->x2 &PAM);;
if(pbox->x1 & PAM){
nlMiddle -= (PAM+1 - (pbox->x1 &PAM));
}
nlMiddle >>= P2WSH;
if(nlMiddle < 0)
nlMiddle = 0;
#endif
#endif
#ifdef DO_MEMCPY
if ((xdir == 1) || (pptSrc->y != pbox->y1)
|| (pptSrc->x + w <= pbox->x1))
{
#if PSZ == 24
char *psrc = (char *) psrcLine + (pptSrc->x * BYPP);
char *pdst = (char *) pdstLine + (pbox->x1 * BYPP);
#else
char *psrc = (char *) psrcLine + pptSrc->x;
char *pdst = (char *) pdstLine + pbox->x1;
#endif
while (h--)
{
#if PSZ == 24
memcpy(pdst, psrc, w2);
#else
memcpy(pdst, psrc, w);
#endif
pdst += widthDst << PWSH;
psrc += widthSrc << PWSH;
}
}
#else
if (xdir == 1)
{
#if PSZ == 24 && MROP == 0
xoffSrc = ( - pptSrc->x) & PAM;
xoffDst = ( - pbox->x1) & PAM;
#if 1
if((int)xoffSrc != (int)xoffDst
|| ((widthDstBytes & PAM) != (widthSrcBytes & PAM) && h > 1))
#else
if(1)
#endif
{
while (h--)
{
register int i, si, sii, di;
for (i = 0, si = pptSrc->x, di = pbox->x1;
i < w;
i++, si++, di++) {
psrc = psrcLine + ((si * BYPP) >> P2WSH);
pdst = pdstLine + ((di * BYPP) >> P2WSH);
sii = (si & 3);
MROP_SOLID24P(psrc, pdst, sii, di);
}
pdstLine += widthDst;
psrcLine += widthSrc;
}
}
else
#endif
{
#if PSZ == 24
#if MROP != 0
xoffSrc = ( - pptSrc->x) & PAM;
xoffDst = ( - pbox->x1) & PAM;
#endif
pdstLine += (pbox->x1 * BYPP) >> P2WSH;
psrcLine += (pptSrc->x * BYPP) >> P2WSH;
#else
xoffSrc = pptSrc->x & PIM;
xoffDst = pbox->x1 & PIM;
pdstLine += (pbox->x1 >> PWSH);
psrcLine += (pptSrc->x >> PWSH);
#endif
#ifdef DO_UNALIGNED_BITBLT
nl = xoffSrc - xoffDst;
psrcLine = (CfbBits *)
(((unsigned char *) psrcLine) + nl);
#else
#if PSZ == 24 && MROP == 0
#else
if (xoffSrc == xoffDst)
#endif
#endif
{
while (h--)
{
#if PSZ == 24 && MROP == 0
register int index;
register int im3;
#endif
psrc = psrcLine;
pdst = pdstLine;
pdstLine += widthDst;
psrcLine += widthSrc;
#if PSZ == 24 && MROP == 0
index = (int)(pdst - pdstBase);
im3 = index % 3;
#endif
if (startmask)
{
#if PSZ == 24 && MROP == 0
*pdst = DoMaskMergeRop24u(*psrc, *pdst, startmask, im3);
index++;
im3 = index % 3;
#else
*pdst = MROP_MASK(*psrc, *pdst, startmask);
#endif
psrc++;
pdst++;
}
nl = nlMiddle;
#ifdef LARGE_INSTRUCTION_CACHE
#ifdef FAST_CONSTANT_OFFSET_MODE
psrc += nl & (UNROLL-1);
pdst += nl & (UNROLL-1);
#if PSZ == 24 && MROP == 0
#define BodyOdd(n) pdst[-n] = DoMergeRop24u(psrc[-n], pdst[-n], ((int)(pdst - n - pdstBase))%3);
#define BodyEven(n) pdst[-n] = DoMergeRop24u(psrc[-n], pdst[-n], ((int)(pdst - n - pdstBase))%3);
#else
#define BodyOdd(n) pdst[-n] = MROP_SOLID (psrc[-n], pdst[-n]);
#define BodyEven(n) pdst[-n] = MROP_SOLID (psrc[-n], pdst[-n]);
#endif
#define LoopReset \
pdst += UNROLL; \
psrc += UNROLL;
#else
#if PSZ == 24 && MROP == 0
#define BodyOdd(n) *pdst = DoMergeRop24u(*psrc, *pdst, im3); pdst++; psrc++; index++; im3 = index % 3;
#define BodyEven(n) BodyOdd(n)
#else
#define BodyOdd(n) *pdst = MROP_SOLID (*psrc, *pdst); pdst++; psrc++;
#define BodyEven(n) BodyOdd(n)
#endif
#define LoopReset ;
#endif
PackedLoop
#undef BodyOdd
#undef BodyEven
#undef LoopReset
#else
#ifdef NOTDEF
while ((nl -= 6) >= 0)
{
asm ("moveml %1+,#0x0c0f;moveml#0x0c0f,%0"
: "=m" (*(char *)pdst)
: "m" (*(char *)psrc)
: "d0", "d1", "d2", "d3",
"a2", "a3");
pdst += 6;
}
nl += 6;
while (nl--)
*pdst++ = *psrc++;
#endif
#if 0
DuffL(nl, label1,
*pdst = DoMergeRop24u(*psrc, *pdst, im3);
pdst++; psrc++; index++;im3 = index % 3;)
#else
DuffL(nl, label1,
*pdst = MROP_SOLID (*psrc, *pdst);
pdst++; psrc++;)
#endif
#endif
if (endmask)
#if PSZ == 24 && MROP == 0
*pdst = DoMaskMergeRop24u(*psrc, *pdst, endmask, (int)(pdst - pdstBase) % 3);
#else
*pdst = MROP_MASK(*psrc, *pdst, endmask);
#endif
}
}
#ifndef DO_UNALIGNED_BITBLT
#if PSZ == 24 && MROP == 0
#else
else
{
if (xoffSrc > xoffDst)
{
#if PSZ == 24
leftShift = (xoffSrc - xoffDst) << 3;
#else
#if PGSZ == 32
leftShift = (xoffSrc - xoffDst) << (5 - PWSH);
#else
leftShift = (xoffSrc - xoffDst) << (6 - PWSH);
#endif
#endif
rightShift = PGSZ - leftShift;
}
else
{
#if PSZ == 24
rightShift = (xoffDst - xoffSrc) << 3;
#else
#if PGSZ == 32
rightShift = (xoffDst - xoffSrc) << (5 - PWSH);
#else
rightShift = (xoffDst - xoffSrc) << (6 - PWSH);
#endif
#endif
leftShift = PGSZ - rightShift;
}
while (h--)
{
psrc = psrcLine;
pdst = pdstLine;
pdstLine += widthDst;
psrcLine += widthSrc;
bits = 0;
if (xoffSrc > xoffDst)
bits = *psrc++;
if (startmask)
{
bits1 = BitLeft(bits,leftShift);
bits = *psrc++;
bits1 |= BitRight(bits,rightShift);
*pdst = MROP_MASK(bits1, *pdst, startmask);
pdst++;
}
nl = nlMiddle;
#ifdef LARGE_INSTRUCTION_CACHE
bits1 = bits;
#ifdef FAST_CONSTANT_OFFSET_MODE
psrc += nl & (UNROLL-1);
pdst += nl & (UNROLL-1);
#define BodyOdd(n) \
bits = psrc[-n]; \
pdst[-n] = MROP_SOLID(BitLeft(bits1, leftShift) | BitRight(bits, rightShift), pdst[-n]);
#define BodyEven(n) \
bits1 = psrc[-n]; \
pdst[-n] = MROP_SOLID(BitLeft(bits, leftShift) | BitRight(bits1, rightShift), pdst[-n]);
#define LoopReset \
pdst += UNROLL; \
psrc += UNROLL;
#else
#define BodyOdd(n) \
bits = *psrc++; \
*pdst = MROP_SOLID(BitLeft(bits1, leftShift) | BitRight(bits, rightShift), *pdst); \
pdst++;
#define BodyEven(n) \
bits1 = *psrc++; \
*pdst = MROP_SOLID(BitLeft(bits, leftShift) | BitRight(bits1, rightShift), *pdst); \
pdst++;
#define LoopReset ;
#endif
PackedLoop
#undef BodyOdd
#undef BodyEven
#undef LoopReset
#else
DuffL (nl,label2,
bits1 = BitLeft(bits, leftShift);
bits = *psrc++;
*pdst = MROP_SOLID (bits1 | BitRight(bits, rightShift), *pdst);
pdst++;
)
#endif
if (endmask)
{
bits1 = BitLeft(bits, leftShift);
if (BitLeft(endmask, rightShift))
{
bits = *psrc;
bits1 |= BitRight(bits, rightShift);
}
*pdst = MROP_MASK (bits1, *pdst, endmask);
}
}
}
#endif
#endif
}
}
#endif
else
{
#if PSZ == 24 && MROP == 0
xoffSrc = (-(pptSrc->x + w)) & PAM;
xoffDst = (-pbox->x2) & PAM;
#if 1
if(xoffSrc != xoffDst
|| ((widthDstBytes & PAM) != (widthSrcBytes & PAM) && h > 1))
#else
if(1)
#endif
{
while (h--)
{
register int i, si, sii, di;
for (i = 0, si = pptSrc->x + w - 1, di = pbox->x2 - 1;
i < w;
i++, si--, di--) {
psrc = psrcLine + ((si * BYPP) >> P2WSH);
pdst = pdstLine + ((di * BYPP) >> P2WSH);
sii = (si & PAM);
MROP_SOLID24P(psrc, pdst, sii, di);
}
psrcLine += widthSrc;
pdstLine += widthDst;
}
}else
#endif
{
#if PSZ == 24
#if MROP == 0
#else
xoffSrc = (pptSrc->x + w) & PAM;
xoffDst = pbox->x2 & PAM;
#endif
pdstLine += ((pbox->x2 * BYPP - 1) >> P2WSH) + 1;
psrcLine += (((pptSrc->x+w) * BYPP - 1) >> P2WSH) + 1;
#else
xoffSrc = (pptSrc->x + w - 1) & PIM;
xoffDst = (pbox->x2 - 1) & PIM;
pdstLine += ((pbox->x2-1) >> PWSH) + 1;
psrcLine += ((pptSrc->x+w - 1) >> PWSH) + 1;
#endif
#ifdef DO_UNALIGNED_BITBLT
#if PSZ == 24
nl = xoffDst - xoffSrc;
#else
nl = xoffSrc - xoffDst;
#endif
psrcLine = (CfbBits *)
(((unsigned char *) psrcLine) + nl);
#else
#if PSZ == 24 && MROP == 0
#else
if (xoffSrc == xoffDst)
#endif
#endif
{
while (h--)
{
#if PSZ == 24 && MROP == 0
register int index;
register int im3;
#endif
psrc = psrcLine;
pdst = pdstLine;
pdstLine += widthDst;
psrcLine += widthSrc;
#if PSZ == 24 && MROP == 0
index = (int)(pdst - pdstBase);
#endif
if (endmask)
{
pdst--;
psrc--;
#if PSZ == 24 && MROP == 0
index--;
im3 = index % 3;
*pdst = DoMaskMergeRop24u(*psrc, *pdst, endmask, im3);
#else
*pdst = MROP_MASK (*psrc, *pdst, endmask);
#endif
}
nl = nlMiddle;
#ifdef LARGE_INSTRUCTION_CACHE
#ifdef FAST_CONSTANT_OFFSET_MODE
psrc -= nl & (UNROLL - 1);
pdst -= nl & (UNROLL - 1);
#if PSZ == 24 && MROP == 0
#define BodyOdd(n) pdst[n-1] = DoMergeRop24u(psrc[n-1], pdst[n-1], ((int)(pdst - (n - 1) -pdstBase)) % 3);
#else
#define BodyOdd(n) pdst[n-1] = MROP_SOLID (psrc[n-1], pdst[n-1]);
#endif
#define BodyEven(n) BodyOdd(n)
#define LoopReset \
pdst -= UNROLL;\
psrc -= UNROLL;
#else
#if PSZ == 24 && MROP == 0
#define BodyOdd(n) --pdst; --psrc; --index; im3 = index % 3;*pdst = DoMergeRop24u(*psrc, *pdst, im3);
#else
#define BodyOdd(n) --pdst; --psrc; *pdst = MROP_SOLID(*psrc, *pdst);
#endif
#define BodyEven(n) BodyOdd(n)
#define LoopReset ;
#endif
PackedLoop
#undef BodyOdd
#undef BodyEven
#undef LoopReset
#else
#if PSZ == 24 && MROP == 0
DuffL(nl,label3,
--pdst; --psrc; --index; im3= index%3;*pdst = DoMergeRop24u(*psrc, *pdst, im3);)
#else
DuffL(nl,label3,
--pdst; --psrc; *pdst = MROP_SOLID (*psrc, *pdst);)
#endif
#endif
if (startmask)
{
--pdst;
--psrc;
#if PSZ == 24 && MROP == 0
*pdst = DoMaskMergeRop24u(*psrc, *pdst, startmask, (int)(pdst - pdstBase) % 3);
#else
*pdst = MROP_MASK(*psrc, *pdst, startmask);
#endif
}
}
}
#ifndef DO_UNALIGNED_BITBLT
#if PSZ == 24 && MROP == 0
#else
else
{
if (xoffDst > xoffSrc)
{
#if PSZ == 24
leftShift = (xoffDst - xoffSrc) << 3;
rightShift = PGSZ - leftShift;
#else
#if PGSZ == 32
rightShift = (xoffDst - xoffSrc) << (5 - PWSH);
#else
rightShift = (xoffDst - xoffSrc) << (6 - PWSH);
#endif
leftShift = PGSZ - rightShift;
#endif
}
else
{
#if PSZ == 24
rightShift = (xoffSrc - xoffDst) << 3;
leftShift = PGSZ - rightShift;
#else
#if PGSZ == 32
leftShift = (xoffSrc - xoffDst) << (5 - PWSH);
#else
leftShift = (xoffSrc - xoffDst) << (6 - PWSH);
#endif
rightShift = PGSZ - leftShift;
#endif
}
while (h--)
{
psrc = psrcLine;
pdst = pdstLine;
pdstLine += widthDst;
psrcLine += widthSrc;
bits = 0;
#if PSZ == 24
if (xoffSrc > xoffDst)
#else
if (xoffDst > xoffSrc)
#endif
bits = *--psrc;
if (endmask)
{
bits1 = BitRight(bits, rightShift);
bits = *--psrc;
bits1 |= BitLeft(bits, leftShift);
pdst--;
*pdst = MROP_MASK(bits1, *pdst, endmask);
}
nl = nlMiddle;
#ifdef LARGE_INSTRUCTION_CACHE
bits1 = bits;
#ifdef FAST_CONSTANT_OFFSET_MODE
psrc -= nl & (UNROLL - 1);
pdst -= nl & (UNROLL - 1);
#define BodyOdd(n) \
bits = psrc[n-1]; \
pdst[n-1] = MROP_SOLID(BitRight(bits1, rightShift) | BitLeft(bits, leftShift),pdst[n-1]);
#define BodyEven(n) \
bits1 = psrc[n-1]; \
pdst[n-1] = MROP_SOLID(BitRight(bits, rightShift) | BitLeft(bits1, leftShift),pdst[n-1]);
#define LoopReset \
pdst -= UNROLL; \
psrc -= UNROLL;
#else
#define BodyOdd(n) \
bits = *--psrc; --pdst; \
*pdst = MROP_SOLID(BitRight(bits1, rightShift) | BitLeft(bits, leftShift),*pdst);
#define BodyEven(n) \
bits1 = *--psrc; --pdst; \
*pdst = MROP_SOLID(BitRight(bits, rightShift) | BitLeft(bits1, leftShift),*pdst);
#define LoopReset ;
#endif
PackedLoop
#undef BodyOdd
#undef BodyEven
#undef LoopReset
#else
DuffL (nl, label4,
bits1 = BitRight(bits, rightShift);
bits = *--psrc;
--pdst;
*pdst = MROP_SOLID(bits1 | BitLeft(bits, leftShift),*pdst);
)
#endif
if (startmask)
{
bits1 = BitRight(bits, rightShift);
if (BitRight (startmask, leftShift))
{
bits = *--psrc;
bits1 |= BitLeft(bits, leftShift);
}
--pdst;
*pdst = MROP_MASK(bits1, *pdst, startmask);
}
}
}
#endif
#endif
}
}
pbox++;
pptSrc++;
}
if (pboxNew2)
{
DEALLOCATE_LOCAL(pptNew2);
DEALLOCATE_LOCAL(pboxNew2);
}
if (pboxNew1)
{
DEALLOCATE_LOCAL(pptNew1);
DEALLOCATE_LOCAL(pboxNew1);
}
}